Historical Plant Data Research Articles

Direct learning of improved control policies from historical plant data

The continuous optimization of the operational performance of chemical plants is of fundamental importance. This research proposes a method that utilizes policy-constrained offline reinforcement learning to learn improved control policies from abundant historical plant data available in industrial settings. As a case study, historical data is generated from a nonlinear chemical system controlled by an economic model predictive controller (EMPC). However, the method’s principles are broadly applicable. Theoretically, it is demonstrated that the learning-based controller inherits stability guarantees from the baseline EMPC. Experimentally, we validate that our method enhances the optimality of the baseline controller while preserving stability, improving the baseline policy by 1% to 20%. The results of this study offer a promising direction for the general improvement of advanced control systems, both data-informed and stability-guaranteed.

Dynamic modeling of Boiler drum using nonlinear system identification approach

Thermal power generation plays a major role in meeting present-day power demands. Efficient and economic controls of crucial elements like steam boiler section need superior control mechanism to ensure better efficiency. Identification of model, dynamics is a formidable task due to the strong coupling of process variables, nonlinearities in the dynamics and constraints on the control inputs. This paper presents the application of nonlinear modeling techniques to estimate boiler drum dynamics in terms of nonlinear model framework which is one of the important requirements for the design of drum water level controls schemes. The historical plant data for estimation is collected from boiler of 210 MW, Stage-II, Boiler-4 of Dr. Narla Tata Rao Thermal power Corporation (NTTPC) Ibrahimpatnam, Vijayawada. Different operating conditions are selected to estimate the nonlinear dynamics of the boiler to check with the existing Hammerstein, Wiener, model frame works. Later the nonlinear framework is modified to estimate other nonlinearities present in the system. The estimated model responses are presented to check the boiler dynamics.

Updating digital twins: Methodology for data accuracy quality control using machine learning techniques

The Digital Twin (DT) constitutes an integration between cyber and physical spaces and has recently become a popular concept in smart manufacturing and Industry 4.0. The related literature provides a DT characterisation and identifies the problem of updating DT models throughout the product life cycle as one of the knowledge gaps. The DT must update its performance by analysing the variable data in real time of the physical asset, whose behaviour is constantly changing over time. The automatic update process involves a data quality problem, i.e., ensuring that the captured values do not come from measurement or provoked errors. In this work, a novel methodology has been proposed to achieve data quality in the interconnection between digital and physical spaces. The methodology is applied to a real case study using the DT of a real solar cooling plant, acting as a learning decision support system that ensures the quality of the data during the update of the DT. The implementation of the methodology integrates a neurofuzzy system to detect failures and a recurrent neural network to predict the size of the errors. Experiments were carried out using historical plant data that showed great results in terms of detection and prediction accuracy, demonstrating the feasibility of applying the methodology in terms of computation time.

The influence of climate warming on flowering phenology in relation to historical annual and seasonal temperatures and plant functional traits.

Climate warming has the potential to influence plant flowering phenology which in turn can have broader ecological consequences. Herbarium collections offer a source of historical plant data that makes possible the ability to document and better understand how warming climate can influence long-term shifts in flowering phenology. We examined the influence of annual, winter, and spring temperatures on the flowering phenology of herbarium specimens for 36 species collected from 1884-2015. We then compared the response to warming between native and non-native, woody and herbaceous, dry and fleshy fruit, and spring vs summer blooming species. Across all species, plants flowered 2.26 days earlier per 1°C increase in annual average temperatures and 2.93 days earlier per 1°C increase in spring onset average temperatures. Winter temperatures did not significantly influence flowering phenology. The relationship of temperature and flowering phenology was not significantly different between native and non-native species. Woody species flowered earlier than herbaceous species only in response to increasing annual temperatures. There was no difference in the phenological response between species with dry fruits and those fleshy fruits for any of the temperature periods. Spring blooming species exhibited a significantly greater phenological response to warming yearly average temperatures than summer blooming species. Although herbarium specimens can reveal climate change impacts on phenology, it is also evident that the phenological responses to warming vary greatly among species due to differences in functional traits such as those considered here, as well as other factors.

Practical Framework for Evaluation and Improvement of Drinking Water Treatment Robustness in Preparation for Extreme-Weather-Related Adverse Water Quality Events.

Robustness is the ability of a drinking water treatment plant (DWTP) to achieve the desired finished water quality even during adverse raw water quality events. Increasing the robustness of a DWTP is beneficial for regular operations and especially for extreme weather adaptation. This paper proposes three robustness frameworks: (a) a general framework outlining the main steps and methodology for systematic assessment and improvement of the robustness of a DWTP, (b) a parameter-specific framework applying the general framework to a water quality parameter (WQP), and (c) a plant-specific framework applying the parameter-specific framework to a DWTP. A parameter-specific framework for turbidity is presented using the turbidity robustness index (TRI) for evaluation and applied to a full-scale DWTP in Ontario, Canada. This evaluation was conducted with historical plant data, as well as bench-scale experimental data simulating extremely high-turbidity scenarios. The framework application is capable of identifying (i) less robust processes which are likely to be vulnerable during climate extremes, (ii) operational responses to increasing short-term robustness, and (iii) a critical WQP threshold beyond which capital improvements are necessary. The proposed framework provides insights into the current state of robustness of a DWTP and serves as a tool for climate adaptation planning.

Data-driven internet of things and cloud computing enabled hydropower plant monitoring system

Hydropower is one of the renewable energy sources that can play a crucial role to fulfil the global energy demand. However, the performance of the hydro turbine is severely affected by silt erosion and cavitation problems which causes a reduction in the overall efficiency of the plant. Various studies have been carried out and are available in the literature to investigate silt erosion and cavitation issues in hydro turbines. It has been reported that cavitation and silt erosion varies with the variation in discharge under part load and overload operating conditions of the machine. However, very few studies are available to predict the performance of the machine under variable operating conditions. Hence, there is a scope of study for monitoring the performance under these conditions in real-time, as it is difficult to predict the behavior of the machine using the existing models. In view of the above, an architecture of a data-driven IoT-based cloud computing-enabled hydropower plant monitoring system has been proposed under the present study. In order to develop this system, historical plant data has been collected and correlations are developed, which are validated with real-time data on the ThingSpeak cloud. It has been found that the developed model can accurately predict the condition of the hydro turbine with an R2-value of 0.9693 having a mean absolute percentage error (MAPE) of 0.67% at 0.89% of root mean square percentage error (RMSPE), and the power factor with an R2-value of 0.9503, having a MAPE of 0.798% at 0.91% of RMSPE.

Integrated Process Re-Design with Operation in the Digital Era: Illustration through an Industrial Case Study

This work discusses what should be the desirable path and correct tools for the optimal re-design and operation of processes in the Industry 4.0 framework, as illustrated in a challenging case study corresponding to a complex network of evaporation plants in a viscose-fiber factory. The goal is to integrate optimal design, to improve the existing cooling systems, together with the optimal operation of the whole network, balancing the initial investment with the potentially achievable savings. A rigorous mathematical model for such optimization purpose has been built. The model explicitly considers different structural alternatives as a superstructure for the incorporation of new equipment into the network. The uncertainty associated to future operating conditions is also considered by using a two-stage stochastic formulation. Furthermore, the model is also the base from which a deterministic real-time optimization (RTO) builds upon to support the daily management of the future network operation. The RTO tool suggests the allocation of different products to evaporation plants, the distribution of the cooling water and the suitable number of heat pumps to switch on for optimal economic operation. Design and operation problems are formulated and solved via mixed-integer non-linear programming and the results have been tested with historical plant data.

Development of a biogas distribution model for a wastewater treatment plant: a mixed integer linear programming approach.

In this paper, we propose a realistic model for gas distribution of an advanced municipal wastewater treatment works and through minimisation of the total cost of gas distribution we perform retrospective optimisation (RO) using historical plant data. This site is the first in the UK with a mixed operational strategy for biomethane produced on site: to burn in combined heat and power (CHP) engines to create electricity, burn in steam boilers for onsite steam use or inject the biomethane into the National Grid. In addition, natural gas can be imported to make up shortfalls in biomethane if required. Implemented using a novel mixed integer linear programming (MILP) approach, to ensure a fast and robust solution, our results indicate the plant operated optimally within accepted tolerance 98% of the time. However, improving plant robustness (such as reducing unexpected breakdown incidents) could yield a significant increase in gas revenue of 7.8%.

Flexibility Analysis of Heat Exchanger Network Retrofit Designs using Monte Carlo Simulation

The aim of this paper is to use Monte Carlo Simulation (MCS) to conduct an analysis of the flexibility and controllability of retrofitted Heat Exchanger Network (HEN) designs. Historical plant data is often uncertain and variable but is an important tool for understanding the real-time operation of a HEN. MCS uses these uncertainties to give an analysis of the retrofitted HENs, and it can be used in a more detailed analysis of the controllability and flexibility. Using Monte Carlo Simulation, inflexibility has been defined as the probability of a HEN to fail to meet temperature targets within defined tolerances. Inflexibility can be used as a tool to guide the retrofit analysis of a HEN. A retrofit HEN design should improve the flexibility of the network without compromising on the expected energy savings. The proposed method is demonstrated using a four-stream HEN problem. The inflexibility of the network was shown to be improved with bypass control and optimal allocation of retrofit area and the proposed method succeeds in augmenting the retrofit design method so that cost-effective, flexible retrofit HENs can be identified.

Transcribing historical plant data into digital format

Transcribing historical plant data into digital format

Energy gap method (EGM) to increase energy efficiency in industrial processes: Successful cases in polymer processing

A novel strategic decision methodology to increase energy efficiency in industrial processes, the Energy Gap Method (EGM), is presented. For this methodology, six different specific energy consumption levels are proposed; four of them are based on historical plant data and measurements in production processes, and the other two levels are based on benchmark references and thermodynamic minimum estimations. Five gaps or differences between specific energy consumptions can be calculated: production, quality, process, technological, and R&D gaps. Actions aiming at the reduction of the largest gaps are prioritized, taking into account the required resources and payback analyses. Four successful industrial case studies applying EGM in polymer processing are presented. In the first study, a blow molding process was diagnosed and the interventions obtained a specific energy consumption (SEC) reduction of 14%. For the second study, the same process used in the first study was employed to intervene in a second machine, obtaining an SEC reduction of 65%. For the third and fourth studies, an automotive part injection molding process was diagnosed and improved, obtaining SEC savings of 16% for the part production and around 7% for the whole production plant. EGM is not limited to polymer processing plants, and it could be adapted to other mass production processes.

Big data analytics for proactive industrial decision support

Big data technologies offer new opportunities for analyzing historical data generated by process plants. The development of new types of operator support systems (OSS) which help the plant operators during operations and in dealing with critical situations is one of these possibilities. The project FEE has the objective to develop such support functions based on big data analytics of historical plant data. In this contribution, we share our first insights and lessons learned in the development of big data applications and outline the approaches and tools that we developed in the course of the project.

Big data analytics for proactive industrial decision support

Big data technologies offer new opportunities for analyzing historical data generated by process plants. The development of new types of operator support systems (OSS) which help the plant operators during operations and in dealing with critical situations is one of these possibilities. The project FEE has the objective to develop such support functions based on big data analytics of historical plant data. In this contribution, we share our first insights and lessons learned in the development of big data applications and outline the approaches and tools that we developed in the course of the project.

Prediction of Effluent Treatment Plant Performance in a Diary Industry Using Artificial Neural Network Technique

Use of Artificial Neural Network (ANN) models is progressively increasingly to predict waste water treatment plant variables. This forecasting helps the operators to take corrective action and manage the process accordingly as per the norms. It is a proved useful device to surmount a few of the limitations of usual mathematical models for wastewater treatment plants for the reason that of their complex mechanisms, changing aspects-dynamics and inconsistency. This analysis considers the relevance of ANN techniques to predict influent and effluent biochemical oxygen demand (BOD), Chemical Oxygen Demand (COD), Total suspended solids (TSS) for effluent treatment process. Here, a feed forward ANN, using a back propagation learning algorithm, has been applied for predicting effluent BOD, COD, TSS. After collecting historical plant data from effluent treatment plant at Diary industry. The suitable architecture of the neural network models was ascertained after several steps of training and testing of the models. Efficiencies of the plant for BOD, COD, TSS were 85%,78%,75% respectively. The ANN based models were established to offer an efficient and a robust tool in prediction and modelling.

Hybrid water treatment cost prediction model for raw water intake optimization

In order to reduce the total cost of a dual source drinking water treatment plant operation, a comprehensive hybrid prediction model was built to estimate the necessary chemicals dosage and pumping energy costs for alternative source selection scenarios. Correlations between the water quality parameters and the required treatment chemicals were estimated using historical data and the expected pH variations associated with each chemical addition, which was based on the Caldwell–Lawrence diagram. The pumping energy costs were also estimated using a data-driven approach that was based on historical plant data. The research has practical implications for water treatment operators seeking to minimize plant operational costs through selecting raw water intake volumes for their treatment plant based on multiple source options and offtake tower gate levels. Future research seeks to better link current and future water treatment dosage cost predictions directly to water quality measurements taken from vertical profiling systems.

Integration options for solar thermal with low temperature industrial heat recovery loops

This paper investigates three general methods for integrating industrial solar heating into an HRL (Heat Recovery Loop) using an industrial case study. Integration of solar thermal energy into low temperature pinch processes, like dairy and food and beverage processes, is more economic when combined with an HRL to form a combined inter-plant heat recovery and renewable energy utility system. The combined system shares common infrastructure such as piping, pumping and storage, and improves solar heat utilisation through direct solar boosting of the HRL intermediate fluid's temperature and enthalpy either through parallel or series integration relative to the other sources and sinks in the HRL system. Three options for integrating solar thermal directly into HRLs are dynamically modelled using historical plant data from a large multi-plant dairy case study to demonstrate the hot utility savings potential of the solar-HRL system. For the dairy case study, results show the best location for integrating solar heating into an HRL is in series with the heat sources as the hot fluid returns to the storage tank. This configuration maximises the effectiveness of collecting solar heat as a meaningful replacement of non-renewable process heat.

Options for Solar Thermal and Heat Recovery Loop Hybrid System Design

Integration of solar thermal energy into low temperature pinch processes, like dairy and food and beverage processes is more economic when combined with a Heat Recovery Loop (HRL) to form a hybrid inter-plant heat recovery system. The hybrid system shares common infrastructure and improves solar heat utilisation through direct solar boosting of the HRL intermediate fluid’s temperature and enthalpy either through parallel or series application. The challenge of dealing with variable solar energy supply is less of a problem in the hybrid system because the HRL with its associated storage acts as an enthalpy buffer which absorbs temperature and flow rate fluctuations on both the heat supply (including solar) and heat demand side simultaneously. Three options for integrating solar thermal directly into HRLs are applied to a large multi-plant dairy case study to demonstrate the hot utility savings potential of the Solar-HRL hybrid system. HRL performance with Variable Temperature Storage (VTS) and solar is dynamically modelled with historical plant data. The series configuration is shown to be consistently better than parallel configuration for the same thermal storage volumes and similar heat exchanger areas.

State of the Art in the Development of Adaptive Soft Sensors based on Just-in-Time Models

Data-driven soft sensors have gained popularity due to availability of the recorded historical plant data. The success stories of the implementations of soft sensors, however, involved some practical difficulties. Even if a good soft sensor is successfully developed, its predictive performance will gradually deteriorate after a certain time due to changes in the state of plants and process characteristics, such as catalyst deactivation and sensor and process drifts due to equipment ageing, fouling, clogging and wear, changes of raw materials and so on. To get soft sensor automatically updated, different kinds of methods have been introduced, such as Kalman filter, moving window average, recursive and ensemble methods. However, these methods have some drawbacks which motivate the development and implementation of just-in-time (JIT) model based adaptive soft sensor. This paper aims to report the current status of adaptive soft sensors based on just-in-time modelling approach. Critical review and discussion on the original and modified algorithms of the JIT modelling approach are presented. Proposed topics for future research and development are also outlined to provide a road map on the developing improved and more practical adaptive soft sensors based on JIT models.

Reactor Modeling and Recipe Optimization of Ring-Opening Polymerization: Block Copolymers

This paper addresses reactor modeling and recipe optimization of semibatch ring-opening polymerization processes for making block copolymers. Two rigorous reactor models are developed on the basis of the population balance and method of moments, respectively. The complete polymerization process model also includes vapor–liquid equilibrium equations from applying Flory–Huggins theory. The accuracies of both reactor models are validated against historical plant data by adjusting model parameters such as kinetic rate constants. The recipe optimization problem is formulated to design the optimal reactor operating policy to minimize polymerization time and incorporate additional process constraints in accordance with final product properties and process safety requirements. The resulting dynamic optimization problem is translated to a nonlinear program by using the simultaneous collocation method, and further solved by the interior point method. In the case study example, both reactor models show satisfactory matches between their predictions and the historical plant data. The recipe optimization with both models demonstrates significant process improvement and reductions in batch operating time. Moreover, the moment model shows superiority over the population balance model in terms of computational efficiency.

Advanced online thermochemical model for metallurgical control at Xstrata Nickel’s Sudbury smelter

An advanced online thermochemical model was developed for Xstrata Nickel’s Sudbury smelter. The model is connected to a PI server to upload plant operating data in real time. The PI server receives and archives plant data through various interfaces, while PI DataLink retrieves the data from the server into a Microsoft Excel spreadsheet. The model was built on the ChemSheet platform, which combines the practicality and simplicity of a spreadsheet, with the robustness of FactSage, a well known thermochemical calculator. The interface of the model is relatively simple and easy-to-use. For effective modelling, the industrial unit operations were divided into thermochemical modules; the modules were then interconnected and compiled sequentially to produce the final result. Online plant data, such as flowrates, compositions and operating conditions, are dynamically entered into the model. For a given feed scenario, the model converges in ∼20 s, producing final results for calcine, matte, slag and off-gas compositions.The smelter primarily treats Ni–Cu sulphide concentrate via roasters, electric furnace and converters, producing a high grade Bessemer matte product for further refining in Norway. The described model integrates the thermochemistry of the roasters and electric furnace, and predicts important process parameters such as degree of sulphur elimination in the fluid-bed roasters, matte grade, iron metallisation, slag losses and the iron to silica ratio in the electric furnace slag. One of the main objectives of the model was to assist process engineers and operators in calculating the addition rates of coke, flux and air for different feed scenarios. The model is also valuable in helping to stabilise the metallurgy of the converter aisle operation, which is a batch process. Before implementation, the model was extensively validated using daily historical plant data for five randomly selected months from years 2011–2012. The predicted values from the model were found to be in good agreement with the plant data.